SCH: AI-empowered wearable multimodal sensors (AI-MEDALLION) for noninvasive monitoring

Over the last decade, significant progress has been made in wearable sensors that detect biomarkers in a continuous and non-invasive manner from biofluids such as sweat. Parkinson's disease (PD) is the second-most common neurodegenerative disorder in the United States, and early diagnosis and management of the disease course of PD remains an urgent task. Among all biomarker types, the sweat

metabolites, peptides, and anions can reflect both the phenotypic states of cells or organs and their dynamic responses to external stimuli, such as drug treatment. However, significant technical challenges exist in developing sweat metabolic biomarkers and wearable sweat sensors in PD monitoring: 1) lack of

known biomarkers and accurate whole metabolomic and peptide profiles in PD patients' sweat; 2) the selected biomarkers need to be optimized for PD system and sensor technology; 3) current wearable sweat sensors fall short in sensitivity. This project aims to fill these gaps by proposing a novel framework

for predicting and optimizing the PD metabolic biomarkers using large-scale multi-omics data, to innovate non-invasive wearable sweat sensor with high sensitivity and specificity. Specifically, the biomarker panels and sweat sensor will be developed by bridging state-of-the-art AI and nanotechnologies to assist

the diagnosis and monitoring of PD, and will be validated clinically. The proposed research will provide an integrative infrastructure for developing novel PD biomarkers and non-invasive personal wearable devices. We will develop computational methods to solve a set of unaddressed questions in biomarker

discovery, namely identification of PD-specific functional variations, prediction of metabolites and peptides in body fluids, and optimization of biomarker selection for personal wearable sensor (Aim 1). We will gain fundamental knowledge and technical capabilities in designing, processing, and optimizing of 2D

materials-based wearable sensors for versatile applications. A physics-based, data-driven protocol will be developed to reveal the fundamental process-structure-property-performance relation for wearable sensors (Aim 2). We will develop new biomarker sets and wearable sweat sensors and test the performance in assisting diagnosis and monitoring of PD, which could be extended to other

biological/disease systems (Aim 3). This proposed study will fill the gaps of disease-specific biomarker prediction and 2D material optimization in developing a clinically applicable non-invasive wearable device.